Deep-Fake-Detection 0.0.4

Deep Fake Detection Project

Deep Fake Detection Package

---

Abstract

This project Deepfakes can distort our perception of the truth and we need to develop a strategy to improve their detection. Deep Fakes are increasingly detrimental to privacy, social security, and democracy. We plan to achieve better accuracy in predicting real and fake videos..

Instructions to Install our Deep Fake Detection Package

1. Install:

pip install Deep-Fake-Detection

2. Download the Finetunned Model Weights

import gdown
ResNextModelURL    = 'https://drive.google.com/uc?id=1siAcM9uTEoLEeqKFxq5h-hjs4NoHS0SU'
XcePtionModelURL   = 'https://drive.google.com/uc?id=1--68J6Ipny937AFjJ_AKjhFXFnvnGecV'
ResNextModel    = 'resnext.pth'
XcePtionModel   = 'xception.pth'

gdown.download(ResNextModelURL, ResNextModel, quiet=False)
gdown.download(XcePtionModelURL, XcePtionModel, quiet=False)

3. Import the DeepFake_Utils from deepfake_detection :

from deepfake_detection import DeepFake_Utils

4. Detect Detect the video is fake or not by Ensemble ResNext and Inception Custom Models :

# Run the Below Function by Input your Video Path to get the outPutVideo with Label Fake on Real on it
DeepFake_Utils.Inference_on_video(OutputVideoPath,InputVideoPath)


# Show the Video
VideoFileClip(OutputVideoPath, audio=False, target_resolution=(300,None)).ipython_display()